Did Social Media Predict The Super Bowl? No.

Although most of my Cambridge-based colleagues don't want to bring it up, last night's Super Bowl was exactly the spectacle we've come to expect from the nation's most-watched event. We saw hundreds of new commercials (some good and many bad), a crazy half-time show (with a random tightrope walker), and one other thing . . . what was that? Oh, yeah, a football game.

In the weeks leading up to the game, I noticed a trend around the game itself. Dozens of blog posts and news articles claiming they could predict the Super Bowl winner using social media. Although most of these were fluff pieces to fill a slow news week and capitalize on the nation's renewed interest in the NFL, my research skepticism kicked into overdrive with some of them. Not to call anyone out directly, but with all of the PR teams sending me press releases about "predicting" the outcome, I just can't let this slide. So, can social media predict the outcome of the Super Bowl? No.

Each of these predictions came from collecting and analyzing social data. Some predictions came from simple metrics like the volume of mentions around one team against the other. A few of the predictions used the sentiment of mentions — such as a positive mention for the Patriots versus a negative mention for the Giants. And some predictions even used influence calculations to understand how different market segments discussed their favorite teams. In the end, this means that some of the predictions were right and some were wrong. But hey, it was a 50/50 shot anyway. Even with coin-flip odds, it seems that more than half were wrong . . . but that actually distracts from my argument, because even if they guessed right, they were wrong to do so.

It seems many of these predictions forgot that there's absolutely no causal relationship between fans' opinions and players' actions. Just because more people tweeted about one team does not mean that team will win. Unless the NFL changes the rules at some point in the future to award the team with the most social-media-active fans extra points, then the fact stands that these two things have nothing to do with each other. Unlike voting shows like American Idol, where fans can influence the outcome, the Super Bowl comes down to factors outside of fans' control. Using general opinion to predict a game winner simply doesn't work.

I'll spare you the rest of this rant and jump to the reason that I'm ranting in the first place: This misuse of social data is bad for the state of social intelligence. Posting these predictions, whether they're right or wrong, gets people thinking that we can ignore data standards when using social media. It cheapens the data. More and more businesses are treating social media as a valuable source of customer data, but when they see that it incorrectly picked the Patriots to win, it makes them question its validity when they really should question the research methodology.

Maybe I'm just in a bad mood because the Pats lost, or maybe it's time that we stop isolating weak metrics to make unrelated predictions. But as a proponent of social intelligence, I'd say instead that it's time we start treating social media like real data, institute real research standards around social data, and begin answering real business questions.

Comments

Mea Culpa

Zach,

Really appreciate your passion and importance you place on being a voice in helping businesses navigate the hype to find answers to real business questions.

Your post is right on target, we'll stay out of the sports prediction business in the future and stay focused on using social intelligence to understand consumers and develop better marketing strategy and communications.

Full Mea Culpa here: http://www.motivequest.com/blog/index.php/mea-culpa/

Thanks,
Brook

It cheapens the data?

Zach,

Sorry about the Pats. As a Babson student, the last super bowl would not rank as one of my more memorable moments in the past calendar year.
Given the current state of social intelligence, with sentiment analysis tools that are rudimentary at best, do you feel that we have a long ways to go before we can convert social conversations into actionable intelligence?

Suds

Value in social data

Hi Suds- thanks for commenting.

Incidentally, my criticism of predicting the Super Bowl has little to do with the technologies out there. Although I am fairly critical of automated sentiment analysis tools, there are certainly technologies that can make social data actionable.

Actually, Brook's comment addresses this well: social intelligence can inform many things, including marketing and communications strategy, by finding customer insights. As much as this has to do with technologies to aid the process - it's all in the person/team interpreting the data and conducting the requisite actions.

Zach

Further reading

Also, check out the link Brook shared - at the bottom of that post there's another link to a post on the predictive powers of social media. It's a good read.